Hi! I have this problem too. My machine is a Dell Latitude D620. The wireless is an Intel PRO/Wireless 3945ABG using the ipw3945 driver.
The laptop has a switch (on the left side, not that it matters), with three positions: off/on/signal. The first two are fixed (i.e. the switch remains there), while the third has a spring which returns it to the on position when released. I think the third position it's used by the Windows WiFi drivers to trigger some sort of "return to favourites" action, but I don't know for sure, I never used Windows on this machine. The switch can be set in the BIOS to control the Bluetooth/WiFi/Cellular function, or a combination thereof. It is currently set to control all of them (i.e., when turned off all wireless functions should stop). The short story is that: - if I boot-up the laptop with the switch on, WiFi works and the Network Manager configures it just right, - if I turn it off the WiFi stops (i.e. "ifconfig -a" doesn't show the adapter anymore) - if I boot with it off, I never get the adapter shown (as above) - if I turn it on after the system is booted it never gets activated. I looked in the device manager, the hardware is still shown after disappearing from the list given by "ifconfig -a". What is rather interesting is that (after a bit of twiddling with the switch) no events concerning the WiFi hardware appear in syslog. (I think it does show something the first time it's turned off, but my syslog is to messy now to be sure. I'll look again after I reboot.) However, it _does_ generate some keyboard events! Feb 8 20:57:46 bogdanb-d620 kernel: [44072.818000] atkbd.c: Unknown key pressed (translated set 2, code 0x88 on isa0060/serio0). Feb 8 20:57:46 bogdanb-d620 kernel: [44072.818000] atkbd.c: Use 'setkeycodes e008 <keycode>' to make it known. Feb 8 20:57:49 bogdanb-d620 kernel: [44075.535000] atkbd.c: Unknown key pressed (translated set 2, code 0x88 on isa0060/serio0). Feb 8 20:57:49 bogdanb-d620 kernel: [44075.535000] atkbd.c: Use 'setkeycodes e008 <keycode>' to make it known. Feb 8 20:57:52 bogdanb-d620 kernel: [44078.099000] atkbd.c: Unknown key pressed (translated set 2, code 0x91 on isa0060/serio0). Feb 8 20:57:52 bogdanb-d620 kernel: [44078.099000] atkbd.c: Use 'setkeycodes e011 <keycode>' to make it known. Feb 8 20:57:52 bogdanb-d620 kernel: [44078.482000] atkbd.c: Unknown key released (translated set 2, code 0x91 on isa0060/serio0). Feb 8 20:57:52 bogdanb-d620 kernel: [44078.482000] atkbd.c: Use 'setkeycodes e011 <keycode>' to make it known. The above is a sequence starting in the on position; I moved the switch to off, back to on, then "clicked" the third ("attention?") position. If you check the time-stamps, this means that moving between on and off triggers the same key-press e008 (0x88 -- not sure what each code means), with no key-release. Clicking the third position generates a key-press/release event pair with code e011/0x91. I think this is very interesting, as it can be used to notify the kernel or the network manager to do whatever it has to do to re-enable the network adapter, as someone noted earlier. We could even use the "click" key to notify the network adapter to do something interesting, like "look for APs". As for how to do that, I don't yet know. After posting this comment, I'll do a couple of reboots and I'll look at what happens in syslog when I play with the switch. By the way, it might be relevant to note that the same switch in involved in bug #63418. Also, the kernel has a way of telling what is the current position of the switch, because it reports it during boot. -- Doesn't reconnect when wireless disabled/re-enabled using kill switch https://launchpad.net/bugs/20667 -- ubuntu-bugs mailing list ubuntu-bugs@lists.ubuntu.com https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs